whole brain standardization or within a mask?

Submitted by MariaS on

Dear Experts,

for the analysis of our ALFF/fALFF and Reho values we use the z-maps within a graymatter mask. However the mask was only applied after calculation and standardization of those values for the group contrasts. Is this an unusual procedure? or should I use only the voxel within the mask for creating the z-maps?

Thanx in advance for feedback

Maria

YAN Chao-Gan

Tue, 04/19/2016 - 00:23

Hi Maria,

If you only interested in the gray matter voxels, you can standardized them within the GM mask. DPABI->Standardization provides such an option.

Best,

Chao-Gan

Hi,

thanx for the info how to do it. I thought more of the rational behind it, would a reviewer comment on it if I just used the whole brain standardization but only presented data within the mask?

Best,

Maria

Hi all,

Following the above comments, I tested using GroupMask90Percent for rsFC standardization (Mean reg + SD div), then displaying the resulting group stats with four different masks for GRF. NB: no mask was used during postpro or group stats. GroupMask90Percent appears to have been the least conservative for my data; some clusters disappeared altogether when I used GreyMask_02_61x73x61, BrainMask_05_61x73x61, and AAL_61x73x61_YCG. Frequently, the clusters identified included large amounts of WM voxels--including when using GreyMask--can someone please help me understand why masking with GreyMask did not remove CSF and WM activation?

Is it the case that the results obtained with GroupMask90Percent as the GRF mask are the most valid among the options I described? I had wanted to use that one for identifying clusters and to get information on cluster size and peak location and intensity, then use GreyMask to display results, but if I do this I lose some of the effects that I wanted to display. An alternative approach I had considered was to multiply GroupMask90Percent by GreyMask, and use the resulting mask for both standardization and display. Can anyone please share reasons in favor or against this approach?

Thanks in advance,

Felix